virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2025-08-15 14:13:16 +02:00
Code Issues Releases Activity

vp9: support non-420 chroma subsampling for profile 1 token decoding.

Browse Source
This commit is contained in:
Ronald S. Bultje
2015-04-24 10:47:56 -04:00
parent 2d0bea4719
commit 6019002f0f
1 changed files with 38 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

69
libavcodec/vp9.c
View File

@@ -2337,10 +2337,10 @@ static void decode_coeffs(AVCodecContext *ctx)
break; break;
} }
#define DECODE_UV_COEF_LOOP(step) \ #define DECODE_UV_COEF_LOOP(step, decode_coeffs_fn) \
for (n = 0, y = 0; y < end_y; y += step) { \ for (n = 0, y = 0; y < end_y; y += step) { \
for (x = 0; x < end_x; x += step, n += step * step) { \ for (x = 0; x < end_x; x += step, n += step * step) { \
res = decode_coeffs_b(&s->c, s->uvblock[pl] + 16 * n, \ res = decode_coeffs_fn(&s->c, s->uvblock[pl] + 16 * n, \
16 * step * step, c, e, p, a[x] + l[y], \ 16 * step * step, c, e, p, a[x] + l[y], \
uvscan, uvnb, uv_band_counts, qmul[1]); \ uvscan, uvnb, uv_band_counts, qmul[1]); \
a[x] = l[y] = !!res; \ a[x] = l[y] = !!res; \
@@ -2355,36 +2355,30 @@ static void decode_coeffs(AVCodecContext *ctx)
p = s->prob.coef[b->uvtx][1 /* uv */][!b->intra]; p = s->prob.coef[b->uvtx][1 /* uv */][!b->intra];
c = s->counts.coef[b->uvtx][1 /* uv */][!b->intra]; c = s->counts.coef[b->uvtx][1 /* uv */][!b->intra];
e = s->counts.eob[b->uvtx][1 /* uv */][!b->intra]; e = s->counts.eob[b->uvtx][1 /* uv */][!b->intra];
w4 >>= 1; w4 >>= s->ss_h;
h4 >>= 1; end_x >>= s->ss_h;
end_x >>= 1; h4 >>= s->ss_v;
end_y >>= 1; end_y >>= s->ss_v;
for (pl = 0; pl < 2; pl++) { for (pl = 0; pl < 2; pl++) {
a = &s->above_uv_nnz_ctx[pl][col]; a = &s->above_uv_nnz_ctx[pl][col << !s->ss_h];
l = &s->left_uv_nnz_ctx[pl][row & 7]; l = &s->left_uv_nnz_ctx[pl][(row & 7) << !s->ss_v];
switch (b->uvtx) { switch (b->uvtx) {
case TX_4X4: case TX_4X4:
DECODE_UV_COEF_LOOP(1); DECODE_UV_COEF_LOOP(1, decode_coeffs_b);
break; break;
case TX_8X8: case TX_8X8:
MERGE_CTX(2, AV_RN16A); MERGE_CTX(2, AV_RN16A);
DECODE_UV_COEF_LOOP(2); DECODE_UV_COEF_LOOP(2, decode_coeffs_b);
SPLAT_CTX(2); SPLAT_CTX(2);
break; break;
case TX_16X16: case TX_16X16:
MERGE_CTX(4, AV_RN32A); MERGE_CTX(4, AV_RN32A);
DECODE_UV_COEF_LOOP(4); DECODE_UV_COEF_LOOP(4, decode_coeffs_b);
SPLAT_CTX(4); SPLAT_CTX(4);
break; break;
case TX_32X32: case TX_32X32:
MERGE_CTX(8, AV_RN64A); MERGE_CTX(8, AV_RN64A);
// a 64x64 (max) uv block can ever only contain 1 tx32x32 block DECODE_UV_COEF_LOOP(8, decode_coeffs_b32);
// so there is no need to loop
res = decode_coeffs_b32(&s->c, s->uvblock[pl],
1024, c, e, p, a[0] + l[0],
uvscan, uvnb, uv_band_counts, qmul[1]);
a[0] = l[0] = !!res;
AV_WN16A(&s->uveob[pl][0], res);
SPLAT_CTX(8); SPLAT_CTX(8);
break; break;
} }
@@ -3031,34 +3025,39 @@ static void decode_b(AVCodecContext *ctx, int row, int col,
case 8: AV_ZERO64(&v); break; \ case 8: AV_ZERO64(&v); break; \
case 16: AV_ZERO128(&v); break; \ case 16: AV_ZERO128(&v); break; \
} }
#define SPLAT_ZERO_YUV(dir, var, off, n) \ #define SPLAT_ZERO_YUV(dir, var, off, n, dir2) \
do { \ do { \
SPLAT_ZERO_CTX(s->dir##_y_##var[off * 2], n * 2); \ SPLAT_ZERO_CTX(s->dir##_y_##var[off * 2], n * 2); \
if (s->ss_##dir2) { \
SPLAT_ZERO_CTX(s->dir##_uv_##var[0][off], n); \ SPLAT_ZERO_CTX(s->dir##_uv_##var[0][off], n); \
SPLAT_ZERO_CTX(s->dir##_uv_##var[1][off], n); \ SPLAT_ZERO_CTX(s->dir##_uv_##var[1][off], n); \
} else { \
SPLAT_ZERO_CTX(s->dir##_uv_##var[0][off * 2], n * 2); \
SPLAT_ZERO_CTX(s->dir##_uv_##var[1][off * 2], n * 2); \
} \
} while (0) } while (0)
switch (w4) { switch (w4) {
case 1: SPLAT_ZERO_YUV(above, nnz_ctx, col, 1); break; case 1: SPLAT_ZERO_YUV(above, nnz_ctx, col, 1, h); break;
case 2: SPLAT_ZERO_YUV(above, nnz_ctx, col, 2); break; case 2: SPLAT_ZERO_YUV(above, nnz_ctx, col, 2, h); break;
case 4: SPLAT_ZERO_YUV(above, nnz_ctx, col, 4); break; case 4: SPLAT_ZERO_YUV(above, nnz_ctx, col, 4, h); break;
case 8: SPLAT_ZERO_YUV(above, nnz_ctx, col, 8); break; case 8: SPLAT_ZERO_YUV(above, nnz_ctx, col, 8, h); break;
} }
switch (h4) { switch (h4) {
case 1: SPLAT_ZERO_YUV(left, nnz_ctx, row7, 1); break; case 1: SPLAT_ZERO_YUV(left, nnz_ctx, row7, 1, v); break;
case 2: SPLAT_ZERO_YUV(left, nnz_ctx, row7, 2); break; case 2: SPLAT_ZERO_YUV(left, nnz_ctx, row7, 2, v); break;
case 4: SPLAT_ZERO_YUV(left, nnz_ctx, row7, 4); break; case 4: SPLAT_ZERO_YUV(left, nnz_ctx, row7, 4, v); break;
case 8: SPLAT_ZERO_YUV(left, nnz_ctx, row7, 8); break; case 8: SPLAT_ZERO_YUV(left, nnz_ctx, row7, 8, v); break;
} }
} }
if (s->pass == 1) { if (s->pass == 1) {
s->b++; s->b++;
s->block += w4 * h4 * 64; s->block += w4 * h4 * 64;
s->uvblock[0] += w4 * h4 * 16; s->uvblock[0] += w4 * h4 * 64 >> (s->ss_h + s->ss_v);
s->uvblock[1] += w4 * h4 * 16; s->uvblock[1] += w4 * h4 * 64 >> (s->ss_h + s->ss_v);
s->eob += 4 * w4 * h4; s->eob += 4 * w4 * h4;
s->uveob[0] += w4 * h4; s->uveob[0] += 4 * w4 * h4 >> (s->ss_h + s->ss_v);
s->uveob[1] += w4 * h4; s->uveob[1] += 4 * w4 * h4 >> (s->ss_h + s->ss_v);
return; return;
} }
@@ -3155,11 +3154,11 @@ static void decode_b(AVCodecContext *ctx, int row, int col,
if (s->pass == 2) { if (s->pass == 2) {
s->b++; s->b++;
s->block += w4 * h4 * 64; s->block += w4 * h4 * 64;
s->uvblock[0] += w4 * h4 * 16; s->uvblock[0] += w4 * h4 * 64 >> (s->ss_v + s->ss_h);
s->uvblock[1] += w4 * h4 * 16; s->uvblock[1] += w4 * h4 * 64 >> (s->ss_v + s->ss_h);
s->eob += 4 * w4 * h4; s->eob += 4 * w4 * h4;
s->uveob[0] += w4 * h4; s->uveob[0] += 4 * w4 * h4 >> (s->ss_v + s->ss_h);
s->uveob[1] += w4 * h4; s->uveob[1] += 4 * w4 * h4 >> (s->ss_v + s->ss_h);
} }
} }